SPY typically trades with a spread of only 0.004%
Top 10 ETFs trade with typical spreads below 0.01%
Some top 1000 ETFs have spreads over 1%

We were recently approached to put some numbers together comparing spreads between micro-cap stocks, small-cap stocks, large-cap stocks, and a select basket of ETFs.  Brian Livingston, author or Muscular Portfolios, was preparing an article for Marketwatch and needed real-world statistics to support his work.

His findings will not be repeated here because you can see his article on Marketwatch. The table to the right shows how these groups differ even at the end of the day when the market is most liquid and spreads are compressed.

Livingston explores this in detail with an emphasis on the extreme opening spreads and the dangers they present to the careless trader.  Regardless of when you trade, if you are trading less liquid securities it is important to be conscious of spreads and to protect your trades and profits as much as possible.

After seeing the variations noted above we wanted to explore spreads among ETFs.  The chart below focuses on 1,000 of the most active ETFs out of the 2,189 we currently track on this site.  It is clear that while many ETFs trade with reasonable spreads, there are a significant number that do not, and it is important for ETF traders to be just as conscious of spreads as it is stock traders.  We often hear it said that the liquidity of the ETF is not important, but what is important is the liquidity of the underlying assets.  That may be true, but obviously only so if you are using limit orders.

The green dots divide the above distribution into equal quartiles with 250 funds in each of the four segments.  The lower-left segment begins with the major funds such as SPY, QQQ, IWM, and GLD.  These and others like them trade with spreads below 0.01%, with the first quartile topping out at 0.04%.  The second quartile goes up to 0.08% and the third up to 0.16%.  That is getting significant, but it is nothing compared to the last quartile where some spreads approach, and even exceed, the 1% mark.

So where can you find spread data?  Just pull up a symbol for any of the ETFs we track and look in the stats box just below the chart.  The image to the right shows the data for QQQ, but spreads are available for any of the actively traded funds that we track on an intra-day basis.

The fund at the top-right of the above distribution is HSPX, a covered call fund by Mirae Global Assets and marketed under the Global X brand.  This fund trades about 1.4 million dollars per day, so not an insignificant amount and better than most of the ETF universe.  Obviously, decent liquidity does not mean a decent spread.  While there is a general relationship between trade volume and spreads, it is not 100%, and spreads can vary.  Also, note that spreads vary significantly during the day, and these numbers are averages across the day from open to close.  You can expect them to be higher at the open and lower near the close.

Pay attention, or pay the spread.

Livingston presents three models to the reader targeted to different portfolio sizes.  The two primary portfolios require only a monthly review and each generally holds three funds.  The difference being that one selects from a larger asset class pool and will likely be more active.  The simplest portfolio is targeted as a starter portfolio and is a simple two fund portfolio with annual re-balancing.  Full historical performance is provided with significant discussion of drawdowns and bear market conditions.  It’s really refreshing to see an author address the down market realities head-on.  There is no running from the fact that markets go down and portfolios do also.

In addition to these models, Livingston pulls together much of the research and learning that has defined portfolio theory over the past generation.  From Modern Portfolio Theory to fractals, it is covered along with quotes and investing tid-bits from the likes of Bill Sharpe, Robert Shiller, and Warren Buffett.

Muscular Portfolios is packed full of charts and other visuals to support the points being made by the author.  One of my favorite charts is documenting Buffett’s performance during bull and bear markets since 2000, emphasizing that he has under performed the market during the bull periods but still beat the averages by a large margin over time.

Other topics covered range from portfolio creation, to selecting a broker, to minimizing tax consequences.  It obviously covers a lot of territory but is well presented with numerous visuals making the book approachable and educational for both the beginner and the experienced investor.

I should point out that Mr. Livingston has been a long time member and supporter of our ETFScreen.com site, and that I’ve had numerous conversations with him over the past several years related to investing in general and the research he has done for this book.  He has been a benefit to me personally, and his feedback has positively impacted the site.

The book is available at both Amazon and Barnes and Noble, and probably at your favorite book seller.  I’ll end with a couple of quotes from people you might recognize:

“I know of no book for a general investment audience that is more thoroughly researched and backed up by hard data.”
—Mark Hulbert, founder of the Hulbert Financial Digest

“Don’t let your portfolio atrophy, read Muscular Portfolios and pump up your wealth.”
—Mebane Faber, coauthor of The Ivy Portfolio

Our goal with this site was to make most information available for “free”, supported by advertising.  That is probably because we’re cheap and don’t like subscriptions, but we also don’t like abusive advertising so we have stressed to our ad network to only show acceptable ads.

We have shared our concern about the increasing use of ad-blocking software in the past, particularly among our non-subscribers..  We have watched our stats as the percentage of users blocking ads has grown from the mid-teens percent to the current mid-thirty percent level.  That trend is only sustainable if those choosing not to view ads are willing to subscribe to our Premium Access service or provide some other economic benefit.

If that weren’t bad enough, we recently ran a study on all pages served and found some disturbing statistics.  First, let us categorize our users into three groups which we’ll call Good Bots, Bad Bots, and Web Users.  The last category, the Web Users, are our target audience.   You and we are generally in that category.  When we speak of 35% of our users not viewing ads,  we’re speaking of 35% of the targeted Web Users.

The term Good Bot is applied liberally here.  It includes the bots, crawlers, and spiders from the likes of Google, Bing, Yahoo, Baidu (China), Yandex (Russia), and a slew of lesser known players. The term Bad Bot is generally reserved for an automated system pretending to be a regular browser.  A few of these may be corporate based, but many are from traders that have automated the harvesting of data from our ETFScreen.com site.   A few even access the  site with a valid user ID.

What we found is that most of the pages we create are at the request of the Bad Bots, utilizing precious server time. We also found that the “Good Bots” actually request more pages than our targeted Web Users do.  Finally, we found that the bills are being paid by roughly 15% of the pages being viewed.

The Way Forward

Most of the “Good Bots” are truly good bots and provide a benefit by making the site known to our target audience.  Some, however, provide little benefit to our particular site, and a few of these have already been restricted from our site.  Others will be restricted in the near future.

The “Bad Bots” are a mixed lot as well.  Many we’re sure are innocently created by skilled individuals who mean no harm.  But as a group they have a significant negative effect and will find the door closing over the coming weeks.  If you have automated access to our site this is your opportunity to reach out and establish a legitimate path forward.  We have developed a token that subscribing users can utilize in such scenarios so we know who is requesting the pages.  Our site is not designed to serve 1000 pages per hour to a single user, as some have been requesting, so please throttle your requests even with a subscription and the use of the token.  If you have high-volume needs please contact us.

That gets us to the Web Users and the fact that some economically support the site and others do not.  We have recently been placing our own banners when ads are blocked, asking for user’s support.  We hope most of the ad-blockers will take the hint, but the reality is that some will not.  The situation has progressed to the breaking point, and we will soon be limiting access to users that do not view ads and do not subscribe. This is the only way to keep the site functioning.

In exchange, we are committed to only showing ad formats that we find acceptable and relatively low impact. If you see a pop-up ad on our site, or an auto starting video [with audio], we want to know about it and are committed to eliminating such.  In the end we want to have a better performing site serving pages effectively to a deserving audience.

We’re asking for your support have flexibility in how we move forward. Discussion is welcome.

Note:  If you utilize an automated system to retrieve information from our site, please contact us about the use of tokens in the URL.

The funds below all have an RSf reading above 90 and increased in RSf from Jan 26 through today.  More screen details are below if you’re interested.

The two top funds are actively managed ARK funds which have had impressive performance recently.  Beyond this there are several natural resource type funds that fit the thesis of an improving economic environment and possible tightening of supply for raw materials.  Of course there are select technology and biotech funds that make the list, but also note the consumer discretionary funds on the list and the banking funds.  The former is also consistent with the thesis of a healthy consumer and the later with the expectation for higher rates.  Possibly these are the investment themes going forward, maybe.

The screen rules are shown here.  It defines a field called _RSf_8 that is the RSf value from Jan 26 (8 days ago).  It filters for funds with higher RSf values today.  It also excludes leveraged funds and sets a minimum dollar-volume to focus the list on the more significant funds.  The PREVAL() function utilized in the screen is available only to our Premium Access users, but any of those that would like to see the actual screen, or make tweaks to it, can find it here.

We are currently releasing the library with support limited to nine popular indicators including two moving averages (EMA and SMA) and seven momentum indicators (ADX, AROON, CCI, MACD, PPO, ROC, and RSI).  Other indicators will be released over the next few days, but these most popular indicators are ready to go.

The process to include an indicator into your screen is the same as defining a user variable.  Just click on the Add/Create button to open the variable definition box and then click the TA-Lib Helper button in the lower right corner.  This action opens some drop-down selectors you can use  to locate your indicator of choice and, if appropriate, select an output option. 

Most indicators like SMA and ROC take a single input series and output a single series.  With these you usually input price and output a moving average  or rate of change.  Other indicators are more complex with multiple inputs and/or multiple outputs.  For example, in the last image above the chosen indicator is MACD.  Since this indicator outputs three data series (the MACD, the MACD Signal, and the MACD Histogram) the user must select which they want, in this case the MACD Histogram.  If the user wants two or more of the outputs they must define a user variable for each since a  user variable is limited to a single data series.  For most indicators that only output a single series, there is no option for an output series.

While you can type a definition into the Define box it is easier and less error prone to click the “Copy to above” link and let the system copy a default rule to the define box.  From there you can customize either the input fields or the various indicator options.  For those indicators that utilize price data just enter the [Price] field like the default.  This will reference closing data for most indicators, but also make available open, high, low, and volume for indicators that need those fields, like Commodity Channel Index (CCI) and a number of others.

Remember to name your user variable something meaningful and then click OK to save the UV to your screen.  After that, you reference the UV just like any other.

It is beyond the scope of this post to define each indicator and the options available, but most are intuitive if you are familiar with the indicator from other systems.  There is some information available on the TA-Lib Home Page, but a better source for the indicators is probably at this Python related site.

Below are a few examples of TA-Lib function codes:

TaLib(EMA, [Price], 20) – A 20-day Exponential Moving Average(EMA) of the closing prices.

TaLib(SMA, [RSf], 10) – A 10-day Simple Moving Average(SMA) of RSf.  Note you can reference any standard data field, or even another user variable as input.

TaLib(MACD, MACDHist, [Price], 12, 26, 9) – Here an MACD is being defined and you will see is requires more inputs that the simple moving averages shown above.  After the indicator (MACD) is specified an output field must be specified, here it is MACDHist for the histogram data.  After that there are the three defining inputs required for an MACD.  The first two are EMA lengths for calculating two moving averages and the third is used for calculate the signal line.  More information on this and other indicators can be found on Investopedia.

TaLib(EMA, [ _EMA1], 5) – An indicator can reference the output of another by referencing the variable name as input.  Here, if EMA1 had previously been defined as TaLib(EMA, [Price], 20), then the above would give you the 5-day moving average of the 20-day moving average.

PREVAL([_EMA1], 21) – The PREVAL function can be utilized to reference the value of an indicator some days in the past.  In this example we are referencing the value of a variable we named _EMA1, 21 days prior.  Utilizing this one can determine whether an indicator is rising or falling over some time period.  In testing we’ve used this in a filter rule to determine if OBV values were rising over the past month.

We recognize a few additions that need to be made to the text box describing the indicator, like including the name and describing what each option field does.  These tweaks will be made in the near future, but please provide us feedback on other changes that need to occur.

We also need feedback on how to present the Volume based indicator values: AD, ADOSC, and OBV.   These indicators are designed for chart reading and the actual numerical values are meaningless.  For the purposes of this site, how should this data be  presented?  We’ve considered moving average relationships, changes in a moving average over time, and even linear regression slope.  We’re confident there are better options out there so we’re asking for help.  In the meantime these indicators are currently unavailable due to related technical issues.

We’re excited about the possibilities of this new library as it opens many doors for our users to explore.  It’s probably obvious, but these customizations are reserved for our Premium Access users.

Enjoy and please provide feedback.

There have been a number of background changes made the past several weeks building up to this as we modified code that was never intended to support customization.  Most of those changes have gone smoothly, but we have also experienced a few issues.

Friday we released a fix for our screen form’s pop-up menus, the ones used for selecting data fields.  Under certain scenarios, including ad-blocking software and certain browsers, these menus were not working consistently or at all in some cases.  If you are still having problems with these menus, please let us know.

Now for some details regarding the use of these functions.

1. They should be used only in the definition of a User Variable, then the user variable should be referenced in the screen rules.  Any attempt to reference one of these functions directly in a filter rule will result in a ‘function not found’ error.
2. The function calls are consistent, and all follow the form of this SMA request: SMA([Price], 5), which would return a 5-day SMA of price.
   

   User Variable definitions of the new functions.

   Of course, any of our standard data fields could be used instead of price, and the number of days can be anything up to a year or more. Sample definitions are shown in this image, including two 50-day moving averages and one 10-day rate-of-change in $Vol-21.

3. The calculation times are mostly insignificant for use with portfolios and symbol lists.  That said, you can expect to wait a minute or two if you specify ‘All ETFs’ or another large grouping.  To help minimize the calculation times we are utilizing caching logic and other tools where appropriate.

At this point there is still some work to do to improve the robustness of these functions, but as they currently exist they will serve most of the needs of our users.  The areas where you might see some errors pop up are if you try to take a moving average of a moving average, or something similar.  We are working on these issues but would be interested in hearing how you are trying to use these functions when you encounter a problem.  We have found through the years that we don’t always visualize exactly how our users will apply a new functionality, so we need your feedback.

We will post more on these functions later, but they are pretty straight forward so we are releasing them today.   We should note that these, like all advanced functions, are limited to Premium Access users.

Have a good week –

Today we are making some foundational changes that we will build on over the coming weeks.  For the most part, you should not notice these changes, but we want to make you aware and open the door for communication if you experience any issues.

There are, however, three changes you will likely notice and need to be aware of.  The most significant change is the addition of a new screening field actPrice, or the actual unadjusted price.  Historically, the Price field has been an unadjusted price that could be used for testing actual price, such as [Price] > 10, or against a moving average, such as [Price] > [SMA-100].  Additionally, with some “magic” behind the scenes, it could be accurately used to test against a historical value such as [Price] > PreVal( [SMA-50], 10), which compares today’s price to the 50-day SMA from 10 days prior.

Adding the flexibility to define custom indicators, which will use prices adjusted for splits and dividends, means a distinction needs be be drawn between the current adjusted price and the current un-adjusted price. To that end, we will support [Price], which is adjusted, and [actPrice] which is not adjusted.  Since all data on the site is (or will be) adjusted for splits and dividends, the major purpose of the [actPrice] field is to compare to a dollar value such as [actPrice] > 10.

At this point both values are the same, but we suggest you use them as noted above since existing screens will be changed over the next few days, and it will no longer be valid to compare [Price] to a dollar value in a backtest.

The second item to note is a change in the screener field menus to support the new field.  With the addition of the [actPrice] field we have combined the Price with the Volume fields into a new Price and Volume related section.  Nothing major here, just be aware of why it has changed.

Note the “Update” button within the stats table.

The third item has to do with screen update functionality.  We are moving the update time to 5:00pm ET as part of the end-of-day update rountine we perform then.  This means you can update your favorite screen in the evening rather than waiting until the next morning – something we know a few of you would like to do.

As part of this change, screens that have been run recently but are not up-to-date will need to be rerun manually by clicking an update button inside the stats box.  This is an extra step if you are wanting the updated stats, but saves a lot of time if you are researching screen results and don’t mind looking at a 2-day old backtest.  If you find a screen you are interested in and want the updated results then just click the update button. It’s easy.

There are a lot of changes behind the scenes that we won’t address, but please be aware, and let us know if you suspect an issue.  Next week we should begin rolling out the indicator customizations beginning with Rate of Change (ROC) and moving averages (SMA and EMA).  If you are curious why ROC, it can be used with the [Price] field to calculate trailing returns.

Please let us know what new indicators you would like to see included.

The screen above represents a backtest of SPY, entering when the Price rises above the 200-day Simple Moving Average and exiting when Price drops below the SMA.  Options include the type of moving average to use (SMA or EMA), the length of the moving average, and what symbols to enter/exit when signals are generated.

The screen below demonstrates many of the options available with this screen.  Here we are basing our moving averages on the ratio of high beta stocks to low beta stocks by referencing the two funds PowerShares S&P 500 High Beta ETF (SPHB) and PowerShares S&P 500 Low Volatility ETF (SPLV).  It is generally held that when high beta stocks are leading, the market is bullish and when low beta are leading, a more conservative approach is in order.  In this model we will compare the 5-day SMA to the 100-day, moving into equal weight S&P(RSP) when the 5-day leads and into Utilities (XLU) when the 5-day lags.

The results for this screen show an improvement in CAGR as well as most risk metrics when compared to SPY.  If one can handle a little more risk, the results get quite interesting when using a leveraged fund like UPRO in the bullish state and maybe SPLV in the bearish state.

Three interactive charts are included with the standard output for each model.  The first is a price and moving average chart followed by an equity chart and finally a percentage drawdown chart.  At this point this page is available to all users, Premium Access or not.  However, we have seen a continued increase in the number of visitors utilizing ad blocking software so with the roll-out of this page we are including a small “reminder screen” that will appear to users not viewing ads and not logged in as Premium Access.   All we ask is that you either allow ads or sign up for our Premium Access.  On our part, we will do our best to deliver only responsible ads that have a minimal impact on your browsing.  If any of you receive this screen in error please let us know, and we will make refinements.

We have identified several possible extensions to this basic functionality including more statistics to guide you to better screen development.  Some of these extensions will be Premium Access, but we are happy to offer this screen to our users initially without restrictions.

Enjoy!

The Portfolio Backtest page is a simple way to test holding a known portfolio of symbols with periodic re-balancing.  On this page you enter each symbol held and its respective weight in the portfolio, select the re-balancing period, and click the Run Model button.  The resulting stats include CAGR, drawdown, Sharpe Ratio, and more.  Also included are interactive charts showing the equity curve and drawdown.

This page resulted from an email we received a few weeks ago, and was a tool I had recently wished for while reviewing some of the robo-trading sites that are available.  Hopefully, this is the first of several simplified and targeted backtesting pages for moving averages and other strategies.  If you have suggestions in this area let us know.  The Portfolio Backtest page is under the “Screeners” tab on the navigation bar.

We also had an inquiry into why some symbols didn’t have correlation data available on the Correlation Page.  It turns out that this relic of past technological limitations was easily overcome with our current server set-up, so now any symbols we follow will be included in the correlation matrix.

As a reminder, you can click any cell in the correlation matrix and view a chart showing how the correlation between those two symbols has varied over a past period.  This phenomenon is interesting with two markets such as S&P500 versus oil.    These correlations are anything but static.  The Correlation Page is under the “Tools” tab on the navigation bar.

The final item for today relates to ad-blocking software.  Technology is constantly changing, and this technology has progressed over the past year from a nuisance to a significant factor.  If you don’t use ad-blocking software or you are a Premium Access subscriber, you don’t need to read past this point. If, however, you are a non-subscribing user running ad-blocking software, we ask you to white-list our site, or otherwise deactivate the ad-blocking while visiting the ETFScreen.com website.  We try to only have responsible advertising with no audio and no pop-up ads.  We need your support on this issue since the majority of our revenue plan is based on advertising.  Our goal is to offer a significant portion of our content without subscription because we think the world is better that way, and because people with small accounts need information too.  Please help us.